Feature Stylization and Domain-aware Contrastive Learning for Domain Generalization
This work addresses the problem of domain shift in machine learning for applications like image classification, offering a novel approach to improve model generalization across unseen domains, though it is incremental in building upon existing stylization and contrastive learning techniques.
The paper tackles domain generalization by proposing a feature stylization method that synthesizes novel domain features while preserving class semantics, and introduces a domain-aware contrastive loss for robustness. It achieves state-of-the-art results on benchmarks like PACS and Office-Home, with concrete performance gains reported in comparisons.
Domain generalization aims to enhance the model robustness against domain shift without accessing the target domain. Since the available source domains for training are limited, recent approaches focus on generating samples of novel domains. Nevertheless, they either struggle with the optimization problem when synthesizing abundant domains or cause the distortion of class semantics. To these ends, we propose a novel domain generalization framework where feature statistics are utilized for stylizing original features to ones with novel domain properties. To preserve class information during stylization, we first decompose features into high and low frequency components. Afterward, we stylize the low frequency components with the novel domain styles sampled from the manipulated statistics, while preserving the shape cues in high frequency ones. As the final step, we re-merge both components to synthesize novel domain features. To enhance domain robustness, we utilize the stylized features to maintain the model consistency in terms of features as well as outputs. We achieve the feature consistency with the proposed domain-aware supervised contrastive loss, which ensures domain invariance while increasing class discriminability. Experimental results demonstrate the effectiveness of the proposed feature stylization and the domain-aware contrastive loss. Through quantitative comparisons, we verify the lead of our method upon existing state-of-the-art methods on two benchmarks, PACS and Office-Home.